Documentation
/ Guide méthodologique
: 
| VII.2. Using minimisation techniques | ||
|---|---|---|
 | Chapter VII. The Calibration module |  |
The theory behind this method may seem very basic as it consists mainly in a point estimation of a correct configuration that could be performed without any underlying uncertainty modelling. This vision is over-simplistic, since numerical optimisation problems are not simple ones, even for mono-criterion optimisation, because of, for instance:
the regularity and possible local minimum of the cost function under consideration;
the size of the input spaces and the possible constraints that can be applied on each and every single input.
This is not even starting discussing the intricate problems that might arise when the calibration is facing identifiability problem. An insight might be seen using for instance evolutionary algorithms. For more methodological aspects on these issues, please refer to Chapter VI.
| |  | |
| Chapter VII. The Calibration module |  | VII.3. Analytical linear Bayesian estimation |